WO2021164349A1 - Procédé et appareil de prédiction de tension artérielle basés sur un signal de photopléthysmographie - Google Patents

Procédé et appareil de prédiction de tension artérielle basés sur un signal de photopléthysmographie Download PDF

Info

Publication number
WO2021164349A1
WO2021164349A1 PCT/CN2020/129636 CN2020129636W WO2021164349A1 WO 2021164349 A1 WO2021164349 A1 WO 2021164349A1 CN 2020129636 W CN2020129636 W CN 2020129636W WO 2021164349 A1 WO2021164349 A1 WO 2021164349A1
Authority
WO
WIPO (PCT)
Prior art keywords
ppg
matrix
generate
wavelet
factor
Prior art date
Application number
PCT/CN2020/129636
Other languages
English (en)
Chinese (zh)
Inventor
王思翰
曹君
Original Assignee
乐普（北京）医疗器械股份有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 乐普（北京）医疗器械股份有限公司 filed Critical 乐普（北京）医疗器械股份有限公司
Publication of WO2021164349A1 publication Critical patent/WO2021164349A1/fr

Links

Images

Classifications

    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/02—Detecting, measuring or recording pulse, heart rate, blood pressure or blood flow; Combined pulse/heart-rate/blood pressure determination; Evaluating a cardiovascular condition not otherwise provided for, e.g. using combinations of techniques provided for in this group with electrocardiography or electroauscultation; Heart catheters for measuring blood pressure
    • A61B5/021—Measuring pressure in heart or blood vessels
    • A61B5/02108—Measuring pressure in heart or blood vessels from analysis of pulse wave characteristics
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
    • A61B5/7235—Details of waveform analysis
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
    • A61B5/7235—Details of waveform analysis
    • A61B5/7253—Details of waveform analysis characterised by using transforms
    • A61B5/726—Details of waveform analysis characterised by using transforms using Wavelet transforms
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
    • A61B5/7235—Details of waveform analysis
    • A61B5/7264—Classification of physiological signals or data, e.g. using neural networks, statistical classifiers, expert systems or fuzzy systems
    • A61B5/7267—Classification of physiological signals or data, e.g. using neural networks, statistical classifiers, expert systems or fuzzy systems involving training the classification device
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
    • A61B5/7271—Specific aspects of physiological measurement analysis
    • A61B5/7275—Determining trends in physiological measurement data; Predicting development of a medical condition based on physiological measurements, e.g. determining a risk factor

Definitions

  • the present invention relates to the technical field of electrophysiological signal processing, in particular to a blood pressure prediction method and device based on a photoplethysmography signal.
  • the heart is the center of human blood circulation.
  • the heart produces blood pressure through regular pulsation, and then supplies blood to the whole body to complete the body's metabolism.
  • Blood pressure is one of the very important physiological signals of the human body.
  • Human blood pressure contains two important values: systolic blood pressure and diastolic blood pressure. Medically, these two quantities are used to judge whether human blood pressure is normal or not. Long-term continuous observation of these two parameters of blood pressure can help people have a clearer understanding of their own heart health.
  • most of the current traditional blood pressure measurement methods use external force upward pressure detection methods such as pressure gauges, which are not only cumbersome to operate, but also easily cause discomfort to the subject, so they cannot be used multiple times to achieve the purpose of continuous monitoring. .
  • the purpose of the present invention is to provide a blood pressure prediction method and device based on the photoplethysmography signal based on the defects of the prior art, and use the photoplethysmography (PPG) equipment to perform non-invasive data on the tester
  • the acquisition solves the problem that the monitor cannot be continuously observed in conventional monitoring; in order to fully obtain effective signal data from the PPG signal, the embodiment of the present invention uses continuous wavelet transform to decompose the PPG signal; in order to realize automatic learning and prediction capabilities
  • the embodiment of the present invention uses a convolutional neural network model with classification and regression function to predict the decomposed signal to obtain the tester’s blood pressure data (diastolic blood pressure, systolic blood pressure); through the embodiment of the present invention, the cumbersome and cumbersome methods of conventional testing methods are avoided.
  • the sense of discomfort has produced an automatic and intelligent data analysis method, so that the application side can conveniently monitor the measured object multiple times.
  • the first aspect of the embodiments of the present invention provides a blood pressure prediction method based on a photovolography signal, the method including:
  • the scaling factor array includes M scaling factors
  • the movement factor array includes N movement factors
  • the M and the N are both integers
  • the PPG signal segment is subjected to signal decomposition processing using continuous wavelet transform to generate a PPG wavelet coefficient matrix [M,N ];
  • the PPG time-frequency three-dimensional tensor [M,N,3] is reconstructed to generate a PPG convolutional three-dimensional tensor [Y,Y] using the bicubic interpolation algorithm ,3];
  • the Y is the input width threshold of the convolutional network;
  • the classification regression model of the convolutional neural network is used to perform classification regression calculation on the PPG convolution three-dimensional tensor [Y, Y, 3] to generate a PPG prediction blood pressure data pair.
  • the matrix elements of the PPG wavelet coefficient matrix [M, N] are complex wavelet coefficients
  • the value range of the matrix elements of the PPG normalization matrix [M, N] is from 0 to 1;
  • the convolutional neural network classification regression model includes: a two-dimensional convolution layer, a maximum pooling layer, a batch normalization layer, an activation layer, an addition layer, a global average pooling layer, a random discarding layer, and a fully connected layer;
  • the PPG predicted blood pressure data pair includes diastolic blood pressure data and systolic blood pressure data.
  • the acquiring the PPG signal of the photoplethysmography method, and segmenting the PPG signal to generate the PPG signal segment specifically includes:
  • the tester uses a PPG signal acquisition device to perform signal acquisition at a preset sampling frequency to generate the PPG signal; segment the PPG signal according to a preset segment duration threshold to generate multiple PPG signal segments.
  • the PPG signal segment is subjected to signal decomposition processing using a continuous wavelet transform method on the PPG signal segment according to the expansion factor of the expansion factor array, the movement factor of the movement factor array, and the wavelet base type to generate PPG wavelet coefficients Matrix [M,N], including:
  • Step 41 Construct a matrix according to the number of rows as the M and the number of columns as the N, generate a temporary PPG wavelet coefficient matrix [M, N], and initialize all matrix elements of the temporary PPG wavelet coefficient matrix [M, N] Is empty;
  • Step 42 Initialize the value of the first index to 1;
  • Step 43 Initialize the value of the second index to 1;
  • Step 44 extracting the scaling factor generating factor a corresponding to the first index from the scaling factor array, and extracting the moving factor generating factor b corresponding to the second index from the moving factor array;
  • Step 45 using the factor a and the factor b as transformation parameters, and using the continuous wavelet transformation formula corresponding to the wavelet base type to perform continuous wavelet transformation calculation on the PPG signal segment to generate wavelet coefficients WT f (a ,b);
  • the wavelet coefficients WT f (a,b) are complex numbers;
  • Step 46 Add the wavelet coefficients WT(a, b) to the temporary PPG wavelet coefficient matrix [M, N] to add data items;
  • Step 47 Add 1 to the second index
  • Step 48 Determine whether the second index is greater than the N, if the second index is greater than the N, go to step 49, and if the second index is less than or equal to the N, go to step 44;
  • Step 49 Add 1 to the first index
  • Step 50 Determine whether the first index is greater than the M, if the first index is greater than the M, go to step 51, and if the first index is less than or equal to the M, go to step 43;
  • Step 51 Set the PPG wavelet coefficient matrix [M, N] as the temporary PPG wavelet coefficient matrix [M, N].
  • the factor a and the factor b are used as transformation parameters, and a continuous wavelet transformation formula corresponding to the wavelet base type is used to perform continuous wavelet transformation calculation on the PPG signal segment to generate wavelet coefficients WT f (a,b), specifically including:
  • the selected wavelet basis expansion and translation function is Wherein, said a is said factor a; said b is said factor b; said Is a standard constant; said e is Euler's number; said H(t) is a unit step function; said t is a time variable;
  • the wavelet-based expansion and contraction translation function ⁇ a,b (t) use the formula for the PPG signal segment Perform continuous wavelet transform calculation to generate the wavelet coefficients WT f (a, b); wherein, the R is a real number; and the f(t) is the PPG signal segment.
  • the PPG wavelet coefficient matrix [M, N] is converted into a real number matrix by modulo the matrix elements, and the converted matrix is subjected to matrix element value normalization processing to generate a PPG normalized matrix [M,N], including:
  • the PPG real number matrix [M, N] performs data item addition operation; the calculation result of the wavelet coefficient modulus is a real number;
  • the RGB color disk matrix is acquired; the PPG time-frequency tensor conversion is performed on the PPG normalization matrix [M, N] according to the RGB color disk matrix to generate a PPG time-frequency three-dimensional tensor [M, N, 3] ], specifically including:
  • RGB color wheel matrix is specifically [X, 3]; the RGB color wheel matrix includes the X color vectors [3]; the X is an integer;
  • the X As the quantization level, divide the range from 0 to 1 into the X data segments; the data segment includes the data segment index and the data segment threshold range; the data segment index ranges from 1 to The X;
  • the color vector [3] is added to the PPG time-frequency three-dimensional tensor [M, N, 3].
  • the method further includes:
  • the first aspect of the embodiments of the present invention provides a blood pressure prediction method based on a photovolographic signal, using a PPG acquisition device to perform non-invasive data collection on a tester, which solves the problem that the monitor cannot be continuously observed in routine monitoring;
  • the embodiment of the present invention uses continuous wavelet transform to decompose the PPG signal; in order to realize the automatic learning and prediction capabilities, the embodiment of the present invention uses a convolutional neural network model with classification and regression function. Decompose the signal for prediction to obtain the tester's blood pressure data (diastolic blood pressure, systolic blood pressure).
  • a second aspect of the embodiments of the present invention provides a device, the device including a memory and a processor, the memory is used to store a program, and the processor is used to execute the first aspect and the methods in each implementation manner of the first aspect.
  • a third aspect of the embodiments of the present invention provides a computer program product containing instructions, which when the computer program product runs on a computer, causes the computer to execute the first aspect and the methods in the implementation manners of the first aspect.
  • a fourth aspect of the embodiments of the present invention provides a computer-readable storage medium on which a computer program is stored.
  • the computer program implements the first aspect and the methods in the first aspect when executed by a processor. .
  • FIG. 1 is a schematic diagram of a blood pressure prediction method based on photovolography signals according to Embodiment 1 of the present invention
  • FIG. 2 is a schematic diagram of a method for generating a wavelet transform time-frequency diagram of a photoplethysmography signal according to the second embodiment of the present invention
  • FIG. 3 is a schematic diagram of the equipment structure of a blood pressure prediction apparatus based on photovolography signals according to Embodiment 3 of the present invention.
  • the PPG signal is a set of signals that uses the light sensor to identify and record the change in light intensity of a specific light source.
  • the blood flow per unit area in the blood vessel changes periodically, and the corresponding blood volume also changes accordingly, resulting in a periodic change trend in the PPG signal reflecting the amount of light absorbed by the blood.
  • a cardiac cycle consists of two time periods: systolic and diastolic; during systole, the heart does work on the blood throughout the body, causing continuous and periodic changes in intravascular pressure and blood flow volume. The absorption of light is the most; when the heart is in diastole, the pressure on the blood vessels is relatively small.
  • the PPG signal waveform reflecting the light energy absorbed by the blood in the blood vessel is composed of two signals: the systolic period signal and the diastolic period signal; the common PPG signal has two peaks, the first one belongs to the systolic period, and the diastolic period signal. The latter belongs to the diastolic period.
  • a feature calculation and regression model that has been trained by batch PPG signals and corresponding measured blood pressure data can be used to obtain the blood pressure prediction value of the current PPG signal.
  • the obtained characteristic data is obtained by using the blood pressure regression calculation method to obtain the regression data as the prediction result.
  • the extraction of valid data on the one hand, we can extract the signal amplitude from the signal time domain as the characteristic value, on the other hand, we can also use the time-frequency conversion of the signal to extract the changing frequency from the signal frequency domain as the characteristic value. In the latter case, it is necessary to perform time-frequency conversion on the signal first, and then extract the eigenvalues according to the conversion result to form a characteristic matrix.
  • the conventional time-frequency conversion method for signals is through Fourier transform. But Fourier transform, because its time-frequency analysis window is a fixed size, it is easy to lose characteristic data for non-stationary signals.
  • the electrophysiological signals such as PPG signals mentioned in this article are all non-stationary signals that are susceptible to interference.
  • Wavelet transform is a kind of time-frequency analysis method, which inherits the idea of Fourier transform, and can also highlight the local characteristics of the signal from its principle.
  • the embodiment of the present invention uses one of the wavelet transforms: continuous wavelet transform to decompose the PPG signal.
  • Continuous wavelet transform provides an important means for signal locality analysis. Compared with short-time Fourier transform, continuous wavelet transform has a window adjustable property, so it has higher analysis ability for non-stationary signals;
  • the wavelet expansion and translation operation performs multi-scale refinement of the signal, which can achieve a higher time resolution in the high-frequency component of the signal, and a higher frequency resolution in the low-frequency component.
  • Continuous wavelet transform has three core parameters: wavelet base, expansion factor and movement factor. Among them, the wavelet base is a wavelet function specifically used for wavelet transformation, the expansion factor is a scale parameter that will transform itself during the wavelet transformation process, and the movement factor is a movement time parameter that will transform itself during the wavelet transformation process.
  • Fig. 1 is a schematic diagram of a blood pressure prediction method based on a photovolography signal according to Embodiment 1 of the present invention. The method mainly includes the following steps:
  • Step 1 Obtain the PPG signal of photoplethysmography, and divide it into segments to generate PPG signal segments;
  • the tester uses a PPG signal acquisition device to perform signal acquisition at a preset sampling frequency to generate the PPG signal; segment the PPG signal according to a preset segment duration threshold to generate multiple PPG signal segments.
  • Step 2 Obtain wavelet base type, scaling factor array and movement factor array
  • the stretch factor array includes M stretch factors; the movement factor array includes N movement factors; M and N are both integers.
  • Step 3 According to the expansion factor of the expansion factor array, the movement factor of the movement factor array and the wavelet base type, the PPG signal segment is processed by continuous wavelet transform to perform signal decomposition processing to generate the PPG wavelet coefficient matrix [M,N];
  • the matrix elements of the PPG wavelet coefficient matrix [M,N] are wavelet coefficients in complex form
  • Step 31 construct a matrix according to the number of rows as the M and the number of columns as the N, generate a temporary PPG wavelet coefficient matrix [M, N], and initialize the temporary PPG wavelet coefficient matrix [M, N] All matrix elements are empty;
  • Step 32 Initialize the value of the first index to 1;
  • Step 33 Initialize the value of the second index to 1;
  • Step 34 extracting the scaling factor generating factor a corresponding to the first index from the scaling factor array, and extracting the moving factor generating factor b corresponding to the second index from the moving factor array;
  • Step 35 using the factor a and the factor b as transformation parameters, and using the continuous wavelet transformation formula corresponding to the wavelet base type to perform continuous wavelet transformation calculation on the PPG signal segment to generate wavelet coefficients WT f (a ,b);
  • wavelet coefficients WT f (a, b) are complex numbers
  • step 351 when the wavelet basis type is generalized Morse wavelet, select the wavelet basis expansion and translation function as
  • Step 352 according to the wavelet base expansion and contraction translation function ⁇ a, b (t), use the formula Performing continuous wavelet transform calculation to generate the wavelet coefficient WT f (a, b);
  • the R is a real number
  • the f(t) is the PPG signal segment
  • Step 36 adding the wavelet coefficients WT(a, b) to the temporary PPG wavelet coefficient matrix [M, N];
  • Step 37 Add 1 to the second index
  • Step 38 Determine whether the second index is greater than the N, if the second index is greater than the N, go to step 39, and if the second index is less than or equal to the N, go to step 34;
  • Step 39 Add 1 to the first index
  • Step 40 Determine whether the first index is greater than the M, if the first index is greater than the M, go to step 41, and if the first index is less than or equal to the M, go to step 33;
  • Step 41 Set the PPG wavelet coefficient matrix [M, N] as the temporary PPG wavelet coefficient matrix [M, N].
  • Step 4 Perform real number matrix conversion on the PPG wavelet coefficient matrix [M, N] by taking the modulus of the matrix elements, and normalize the matrix element values of the converted matrix to generate the PPG normalization matrix [M, N ];
  • the value range of the matrix elements of the PPG normalization matrix [M,N] is from 0 to 1;
  • step 42 constructing a matrix according to the number of rows as the M and the number of columns as the N, generating a PPG real number matrix [M, N], and initializing all the matrix elements of the PPG real number matrix [M, N] as null;
  • Step 43 sequentially extract the matrix elements of the PPG wavelet coefficient matrix [M, N] to generate temporary wavelet coefficients, perform complex modulus calculation on the temporary wavelet coefficients to generate a wavelet coefficient modulus calculation result, and calculate the wavelet coefficient modulus As a result, add data items to the PPG real number matrix [M, N];
  • the calculation result of the wavelet coefficient modulus is a real number
  • Step 44 Perform normalization processing on the values of all matrix elements of the PPG real number matrix [M, N] to generate the PPG normalization matrix [M, N].
  • Step 5 Obtain the RGB color wheel matrix, and perform PPG time-frequency tensor conversion on the PPG normalized matrix [M, N] according to the RGB color wheel matrix to generate the PPG time-frequency three-dimensional tensor [M, N, 3];
  • Step 51 Obtain the RGB color wheel matrix
  • the RGB color wheel matrix is specifically [X, 3]; the RGB color wheel matrix includes the X color vectors [3]; and the X is an integer;
  • the RGB color model is a color standard in the industry. It obtains a variety of colors by changing the three color channels of red (R), green (G), and blue (B) and superimposing them with each other. Yes, this standard includes almost all colors that human vision can perceive, and is one of the most widely used color systems; assuming that the RGB color wheel matrix includes 256 color vectors, the length of each color vector is 3, including three The numerical value of each primary color; assuming that X is equal to 256, the RGB color wheel matrix includes 256 colors;
  • Step 52 Construct a matrix according to the number of rows as the M and the number of columns as the N, generate a temporary level matrix [M, N], and initialize all matrix elements of the temporary level matrix [M, N] to be empty;
  • Step 53 Using the X as the quantization level, divide 0 to 1 equally into the X data segments; the data segment includes the data segment index and the data segment threshold range; the value of the data segment index From 1 to the X;
  • Step 54 Extract the matrix elements of the PPG normalization matrix [M, N] in turn to generate the first current element, and use the value of the first current element to poll and compare the data segment threshold ranges of all data segments. When the value of the first current element is within the compared data segment threshold range, extract the currently compared data segment index to the temporary level matrix [M, N] to perform a data item addition operation;
  • X is 256
  • 256 it is divided into 256 data segments evenly from 0 to 1
  • 0-1/256 is the first segment
  • 1/256 to 2/256 is the second segment
  • 255/256 to 1 is the 256th segment
  • the value of an element is 1/257, then it belongs to the first segment, then this element
  • the level of should be 1, that is, the value of the element corresponding to this element in the temporary level matrix [M,N] is 1;
  • Step 55 Initialize all matrix elements of the PPG time-frequency three-dimensional tensor [M, N, 3] to be empty;
  • Step 56 Extract the matrix elements of the temporary level matrix [M, N] in sequence to generate a second current element, and extract the corresponding color vector from the RGB color wheel matrix using the value of the second current element as an index [3 ] Generate the current color vector [3], and add the current color vector [3] to the PPG time-frequency three-dimensional tensor [M, N, 3].
  • all element values in the temporary level matrix [M,N] are integers from 1-256. According to the value, a corresponding color point can be extracted from the RGB color wheel matrix, and then the color point can be used as a dimension supplement
  • the one-dimensional vector of the temporary level matrix [M,N] is subjected to matrix upscaling processing to generate the PPG time-frequency three-dimensional tensor [M,N,3], and the actual PPG time-frequency three-dimensional tensor [M,N,3] is derived from A three-dimensional tensor composed of M*N color points.
  • the original signal is decomposed into a two-dimensional complex matrix containing wavelet coefficients, where each row corresponds to a single expansion factor (scale factor), that is, the frequency band obtained by dividing the specified octave; Subsequently, the wavelet coefficients are quantized.
  • the specific process is to perform a modulo operation on each element of the complex matrix, and normalize the real matrix obtained by taking the modulus, and finally obtain a matrix with the value range of the element; then the matrix
  • the elements are mapped to a two-dimensional plane, mapped to a three-dimensional RGB color value through a prescribed color space, and the size of the picture is adjusted to adapt to the input of the convolutional neural network.
  • Step 6 According to the preset input width threshold of the convolutional network, use the bicubic interpolation algorithm to perform tensor shape reconstruction operation on the PPG time-frequency 3D tensor [M,N,3] to generate the PPG convolution 3D tensor [Y, Y,3];
  • Y is the input width threshold of the convolutional network.
  • the size of the PPG time-frequency three-dimensional tensor [M,N,3] is deviated from the input size requirements of the convolutional neural network classification regression model.
  • the PPG time-frequency three-dimensional tensor [M,N,3] If the size of is too small, use the bicubic interpolation algorithm to increase the intermediate value points to achieve the effect of changing the shape of the three-dimensional tensor, and finally generate a PPG convolution three-dimensional tensor [Y,Y,3] that meets the requirements.
  • Step 7 use the convolutional neural network classification regression model to perform classification regression calculation on the PPG convolution three-dimensional tensor [Y, Y, 3] to generate a PPG prediction blood pressure data pair;
  • the PPG predicted blood pressure data pair includes diastolic blood pressure data and systolic blood pressure data.
  • the convolutional network used is a customized convolutional network structure.
  • the convolutional neural network classification regression model includes: two-dimensional convolutional layer, maximum pooling layer, batch normalization layer, activation layer, and addition layer , Global average pooling layer, random discarding layer, and fully connected layer. Through the modification of the network structure, a regression model that can output diastolic and systolic blood pressure at the same time can finally be realized.
  • Fig. 2 is a schematic diagram of a wavelet transform time-frequency diagram generation method of a photovolography signal provided by the second embodiment of the present invention. The method mainly includes the following steps:
  • Step 101 the tester uses the photoplethysmography PPG signal acquisition device to perform signal acquisition at a preset sampling frequency to generate a PPG signal; segment the PPG signal according to a preset segment duration threshold to generate multiple PPG signal segments.
  • Step 102 Obtain the wavelet base type, the stretch factor array, and the movement factor array;
  • the stretch factor array includes M stretch factors; the movement factor array includes N movement factors; M and N are both integers.
  • Step 103 according to the expansion factor of the expansion factor array, the movement factor of the movement factor array, and the wavelet base type, perform signal decomposition processing on the PPG signal segment using continuous wavelet transform to generate a PPG wavelet coefficient matrix [M, N];
  • the matrix elements of the PPG wavelet coefficient matrix [M,N] are wavelet coefficients in complex form
  • Step 1031 construct a matrix according to the number of rows as M and the number of columns as N, generate a temporary PPG wavelet coefficient matrix [M, N], and initialize all matrix elements of the temporary PPG wavelet coefficient matrix [M, N] to be empty;
  • Step 1032 Initialize the value of the first index to 1;
  • Step 1033 Initialize the value of the second index to 1;
  • Step 1034 Extract the scaling factor generation factor a corresponding to the first index from the scaling factor array, and extract the movement factor generation factor b corresponding to the second index from the movement factor array;
  • Step 1035 Using factor a and factor b as transformation parameters, using the continuous wavelet transformation formula corresponding to the wavelet base type, perform continuous wavelet transformation calculation on the PPG signal segment to generate wavelet coefficients WT f (a, b); wavelet coefficients WT f (a,b) are plural;
  • the wavelet basis type is the generalized Morse wavelet
  • select the wavelet basis expansion and translation function as Among them, a is factor a; b is factor b; Is the standard constant; e is Euler's number; H(t) is the unit step function; t is the time variable;
  • wavelet base expansion and contraction function ⁇ a,b (t) use the formula for the PPG signal segment Perform continuous wavelet transform calculation to generate wavelet coefficients WT f (a, b); where R is a real number; f(t) is a PPG signal segment;
  • Step 1036 Add the wavelet coefficients WT (a, b) to the temporary PPG wavelet coefficient matrix [M, N] to add data items;
  • Step 1037 add 1 to the second index
  • Step 1038 determine whether the second index is greater than N, if the second index is greater than N, go to step 1039, if the second index is less than or equal to N, go to step 1034;
  • Step 1039 add 1 to the first index
  • Step 1040 Determine whether the first index is greater than M, if the first index is greater than M, go to step 1041, if the first index is less than or equal to M, go to step 1033;
  • Step 1041 Set the PPG wavelet coefficient matrix [M, N] as a temporary PPG wavelet coefficient matrix [M, N].
  • Step 104 Perform real number matrix conversion on the PPG wavelet coefficient matrix [M, N] by modulo the matrix elements, and perform numerical normalization processing on the converted matrix to generate a PPG normalization matrix [M, N];
  • the value range of the matrix elements of the PPG normalization matrix [M, N] is from 0 to 1.
  • Step 105 Obtain the RGB color wheel matrix, and perform PPG time-frequency tensor conversion on the PPG normalized matrix [M, N] according to the RGB color wheel matrix to generate the PPG time-frequency three-dimensional tensor [M, N, 3].
  • the RGB color model is a color standard in the industry. It obtains a variety of colors by changing the three color channels of red (R), green (G), and blue (B) and superimposing them with each other. Yes, this standard includes almost all colors that human vision can perceive, and is one of the most widely used color systems; assuming that the RGB color wheel matrix includes 256 color vectors, the length of each color vector is 3, including three The numerical value of each primary color; assuming that X is equal to 256, the RGB color wheel matrix includes 256 colors;
  • All element values in the temporary level matrix [M, N] are integers from 1-256. According to the value, a corresponding color point can be extracted from the RGB color wheel matrix, and then the color point can be used as a dimension supplementary one-dimensional
  • the vector performs matrix upgrade processing on the temporary level matrix [M,N] to generate the PPG time-frequency three-dimensional tensor [M,N,3], and the actual PPG time-frequency three-dimensional tensor [M,N,3] is M*N A three-dimensional tensor composed of four color points.
  • the original signal is decomposed into a two-dimensional complex matrix containing wavelet coefficients, where each row corresponds to a single expansion factor (scale factor), that is, the frequency band obtained by dividing the specified octave; Subsequently, the wavelet coefficients are quantized.
  • the specific process is to perform a modulo operation on each element of the complex matrix, and normalize the real matrix obtained by taking the modulus, and finally obtain a matrix with the value range of the element; then the matrix
  • the elements are mapped to a two-dimensional plane, mapped to a three-dimensional RGB color value through a prescribed color space, and the size of the picture is adjusted to adapt to the input of the convolutional neural network.
  • Step 106 Perform image conversion on the PPG time-frequency three-dimensional tensor [M, N, 3] to generate PPG time-frequency map data.
  • the bicubic interpolation algorithm can be used to add pixels between points to achieve the effect of magnifying the image.
  • the bicubic interpolation method is to expand the surrounding 4*4 pixels based on a certain original pixel.
  • the bicubic interpolation algorithm can also be used for abbreviation.
  • the PPG time-frequency three-dimensional tensor [M,N,3] is [224,128,3], indicating that the original image is a bitmap with a size of 224*128.
  • FIG. 3 is a schematic diagram of a device structure of a blood pressure prediction apparatus based on a photovolography signal according to Embodiment 3 of the present invention.
  • the device includes a processor and a memory.
  • the memory can be connected to the processor through a bus.
  • the memory may be a non-volatile memory, such as a hard disk drive and a flash memory, and a software program and a device driver program are stored in the memory.
  • the software program can execute various functions of the foregoing method provided by the embodiments of the present invention; the device driver may be a network and interface driver.
  • the processor is used to execute a software program, and when the software program is executed, the method provided in the embodiment of the present invention can be implemented.
  • the embodiment of the present invention also provides a computer-readable storage medium.
  • a computer program is stored on the computer-readable storage medium, and when the computer program is executed by a processor, the method provided in the embodiment of the present invention can be implemented.
  • the embodiment of the present invention also provides a computer program product containing instructions.
  • the processor is caused to execute the above method.
  • the embodiment of the present invention provides a blood pressure prediction method and device based on a photovolography signal, using a PPG acquisition device to perform non-invasive data collection on a tester, which solves the problem that the monitor cannot be continuously observed in routine monitoring; Fully obtain effective signal data from the PPG signal.
  • the embodiment of the present invention uses continuous wavelet transform to decompose the PPG signal; in order to achieve automatic learning and prediction capabilities, the embodiment of the present invention uses a convolutional neural network model with classification and regression function to decompose the PPG signal.
  • the signal is predicted to obtain the tester’s blood pressure data (diastolic blood pressure, systolic blood pressure); the embodiment of the present invention not only avoids the tediousness and discomfort of conventional testing methods, but also produces an automatic and intelligent data analysis method, so that The application side can conveniently carry out multiple continuous monitoring of the measured object.
  • the steps of the method or algorithm described in combination with the embodiments disclosed herein can be implemented by hardware, a software module executed by a processor, or a combination of the two.
  • the software module can be placed in random access memory (RAM), internal memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disks, removable disks, CD-ROMs, or all areas in the technical field. Any other known storage media.

Landscapes

  • Health & Medical Sciences (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Artificial Intelligence (AREA)
  • Public Health (AREA)
  • Animal Behavior & Ethology (AREA)
  • Veterinary Medicine (AREA)
  • Physiology (AREA)
  • Biophysics (AREA)
  • Pathology (AREA)
  • Biomedical Technology (AREA)
  • Heart & Thoracic Surgery (AREA)
  • Medical Informatics (AREA)
  • Molecular Biology (AREA)
  • Surgery (AREA)
  • General Health & Medical Sciences (AREA)
  • Signal Processing (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Psychiatry (AREA)
  • Cardiology (AREA)
  • Vascular Medicine (AREA)
  • Evolutionary Computation (AREA)
  • Fuzzy Systems (AREA)
  • Mathematical Physics (AREA)
  • Image Processing (AREA)
  • Measuring Pulse, Heart Rate, Blood Pressure Or Blood Flow (AREA)

Abstract

Procédé et appareil de prédiction de tension artérielle basés sur un signal de photopléthysmographie. Le procédé comprend les étapes consistant à : générer des segments de signal PPG (1) ; en fonction d'un type de base d'ondelettes acquis, de facteurs de contraction-expansion et de facteurs de déplacement, réaliser une décomposition de signal sur les segments de signal PPG à l'aide d'un moyen de transformation en ondelettes continues, de façon à générer une matrice de coefficients d'ondelettes PPG (3) ; convertir la matrice de coefficients d'ondelettes PPG en réalisant une opération modulo sur des éléments de la matrice et réaliser un traitement de normalisation sur la matrice convertie pour générer une matrice PPG normalisée (4) ; acquérir une matrice de cercle chromatique RVB et, en fonction de la matrice de cercle chromatique RVB, réaliser une conversion de tenseur sur la matrice PPG normalisée afin de générer un tenseur tridimensionnel temps-fréquence PPG (5) ; en fonction d'un seuil de largeur d'entrée de réseau convolutif, réaliser une reconstruction de forme de tenseur sur le tenseur tridimensionnel temps-fréquence PPG grâce à un algorithme d'interpolation bicubique, de façon à générer un tenseur tridimensionnel convolutif PPG (6) ; et réaliser, à l'aide d'un modèle de classification et de régression de réseaux de neurones convolutifs, une classification et un calcul de régression sur le tenseur tridimensionnel convolutif PPG pour générer une paire de données de tension artérielle prédite par PPG (7).
PCT/CN2020/129636 2020-02-21 2020-11-18 Procédé et appareil de prédiction de tension artérielle basés sur un signal de photopléthysmographie WO2021164349A1 (fr)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202010110279.4 2020-02-21
CN202010110279.4A CN111248880B (zh) 2020-02-21 2020-02-21 一种基于光体积变化描记图法信号的血压预测方法和装置

Publications (1)

Publication Number Publication Date
WO2021164349A1 true WO2021164349A1 (fr) 2021-08-26

Family

ID=70949401

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2020/129636 WO2021164349A1 (fr) 2020-02-21 2020-11-18 Procédé et appareil de prédiction de tension artérielle basés sur un signal de photopléthysmographie

Country Status (2)

Country Link
CN (1) CN111248880B (fr)
WO (1) WO2021164349A1 (fr)

Families Citing this family (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111248880B (zh) * 2020-02-21 2022-07-29 乐普（北京）医疗器械股份有限公司 一种基于光体积变化描记图法信号的血压预测方法和装置
CN112022126A (zh) * 2020-09-28 2020-12-04 无锡博智芯科技有限公司 一种基于CNN-BiLSTM模型和PPG的智能血压预测方法
CN112022125A (zh) * 2020-09-28 2020-12-04 无锡博智芯科技有限公司 一种基于CNN-BiGRU模型和PPG的智能血压预测方法
CN112070067B (zh) * 2020-10-12 2023-11-21 乐普（北京）医疗器械股份有限公司 一种光体积描计信号的散点图分类方法和装置

Citations (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20110077484A1 (en) * 2009-09-30 2011-03-31 Nellcor Puritan Bennett Ireland Systems And Methods For Identifying Non-Corrupted Signal Segments For Use In Determining Physiological Parameters
CN109009033A (zh) * 2018-07-10 2018-12-18 华南理工大学 一种基于小波分析和回声状态网络的血压预测方法
CN109726751A (zh) * 2018-12-21 2019-05-07 北京工业大学 基于深度卷积神经网络识别脑电成像图的方法
CN109872817A (zh) * 2019-02-28 2019-06-11 泉州师范学院 一种基于多因素线索网络的血压预测方法
CN110135510A (zh) * 2019-05-22 2019-08-16 电子科技大学中山学院 一种动态领域自适应方法、设备及计算机可读存储介质
US20190328243A1 (en) * 2016-12-21 2019-10-31 Emory University Methods and Systems for Determining Abnormal Cardiac Activity
CN111248880A (zh) * 2020-02-21 2020-06-09 乐普（北京）医疗器械股份有限公司 一种基于光体积变化描记图法信号的血压预测方法和装置
CN111358455A (zh) * 2020-03-17 2020-07-03 乐普（北京）医疗器械股份有限公司 一种多数据源的血压预测方法和装置

Family Cites Families (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107341769A (zh) * 2016-05-03 2017-11-10 中国科学院微电子研究所 一种心电信号除噪方法及系统
CN108491863A (zh) * 2018-02-27 2018-09-04 南京信息工程大学 基于非负矩阵分解和卷积神经网络的彩色图像处理方法
CN109589101B (zh) * 2019-01-16 2020-08-21 四川大学 一种基于视频的非接触式生理参数获取方法与装置
CN110367950B (zh) * 2019-07-22 2022-06-07 西安奇点融合信息科技有限公司 非接触式生理信息检测方法及系统

Patent Citations (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20110077484A1 (en) * 2009-09-30 2011-03-31 Nellcor Puritan Bennett Ireland Systems And Methods For Identifying Non-Corrupted Signal Segments For Use In Determining Physiological Parameters
US20190328243A1 (en) * 2016-12-21 2019-10-31 Emory University Methods and Systems for Determining Abnormal Cardiac Activity
CN109009033A (zh) * 2018-07-10 2018-12-18 华南理工大学 一种基于小波分析和回声状态网络的血压预测方法
CN109726751A (zh) * 2018-12-21 2019-05-07 北京工业大学 基于深度卷积神经网络识别脑电成像图的方法
CN109872817A (zh) * 2019-02-28 2019-06-11 泉州师范学院 一种基于多因素线索网络的血压预测方法
CN110135510A (zh) * 2019-05-22 2019-08-16 电子科技大学中山学院 一种动态领域自适应方法、设备及计算机可读存储介质
CN111248880A (zh) * 2020-02-21 2020-06-09 乐普（北京）医疗器械股份有限公司 一种基于光体积变化描记图法信号的血压预测方法和装置
CN111358455A (zh) * 2020-03-17 2020-07-03 乐普（北京）医疗器械股份有限公司 一种多数据源的血压预测方法和装置

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
WU, YUDONG ET AL.: "Blood Pressure Measurement Based on Photoelectric Volume Pulse Wave", JOURNAL OF MECHANICAL & ELECTRICAL ENGINEERING, vol. 34, no. 8, 31 August 2017 (2017-08-31), XP009529924, ISSN: 1001-4551 *

Also Published As

Publication number Publication date
CN111248880A (zh) 2020-06-09
CN111248880B (zh) 2022-07-29

Similar Documents

Publication Publication Date Title
WO2021164349A1 (fr) Procédé et appareil de prédiction de tension artérielle basés sur un signal de photopléthysmographie
Lahoud et al. Zero-learning fast medical image fusion
WO2021184805A1 (fr) Procédé et dispositif de prédiction de pression sanguine utilisant une pluralité de sources de données
US8360986B2 (en) Non-contact and passive measurement of arterial pulse through thermal IR imaging, and analysis of thermal IR imagery
US20230355188A1 (en) Systems and methods for image data acquisition
EP2303110B1 (fr) Scalogrammes enchaînés
CN112914527B (zh) 一种基于脉搏波光电容积描记术的动脉血压信号采集方法
TW202008211A (zh) 一種圖像處理方法、電子設備及儲存介質
WO2021164345A1 (fr) Procédé et dispositif de prédiction de pression artérielle
CN109948396B (zh) 一种心拍分类方法、心拍分类装置及电子设备
CN111588367A (zh) 心率检测方法、装置及计算机可读存储介质
CN112806977B (zh) 基于多尺度融合网络的生理参数测量方法
CN112788200B (zh) 频谱信息的确定方法及装置、存储介质、电子装置
Mir et al. ECG denoising and feature extraction techniques–a review
Chekmenev et al. Multiresolution approach for noncontact measurements of arterial pulse using thermal imaging
CN113116321A (zh) 基于pso-grnn神经网络的无创连续血压测量系统
Qayyum et al. Assessment of physiological states from contactless face video: a sparse representation approach
CN113180623A (zh) 无袖式血压测量方法、系统、设备及存储介质
CN117357080A (zh) 近红外光谱信号去噪方法及装置、终端设备、存储介质
Das et al. Time-Frequency Learning Framework for rPPG Signal Estimation Using Scalogram Based Feature Map of Facial Video Data
CN116649924A (zh) 一种脉搏分析方法及装置
CN114145725B (zh) 一种基于无创连续血压测量的ppg采样率估算方法
CN113576422B (zh) 一种生理信号波形的简化处理和量化方法、系统及设备
CN114403831B (zh) 一种ppg波形脉冲提取方法及装置
EP4111953A1 (fr) Procédé et dispositif de traduction entre signaux stochastiques

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 20919924

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 20919924

Country of ref document: EP

Kind code of ref document: A1